MrNeRF (@janusch_patas)

2025-12-04 | โค๏ธ 130 | ๐Ÿ” 14


Radiance Meshes for Volumetric Reconstruction

TL;DR: First method to rasterize a radiance field (not primitives like 3DGS)

Abstract: We introduce radiance meshes, a technique for representing radiance fields with constant density tetrahedral cells produced through Delaunay tetrahedralization. Unlike a Voronoi diagram, Delaunay tetrahedralization yields simple triangles natively supported by existing hardware. As such, our model performs exact and fast volume rendering using both rasterization and ray-tracing.

We introduce a new rasterization method that achieves faster rendering speeds than all prior radiance field representations (assuming an equivalent number of primitives and resolution) across a variety of platforms. Optimizing the positions of Delaunay vertices introduces topological discontinuities (edge flips). To solve this, we use a Zip-NeRF-style backbone, allowing us to express a smoothly varying field even when the topology changes.

Our rendering method exactly evaluates the volume rendering equation, enabling high-quality, real-time view synthesis on standard consumer hardware. Our tetrahedral meshes also lend themselves to a variety of exciting applications, including fisheye lens distortion, physics-based simulation, editing, and mesh extraction.

๐Ÿ”— ์›๋ณธ ๋งํฌ

๋ฏธ๋””์–ด

video thumbnail


Tags

3D AI-ML Simulation